gemini-interactions-api

Pass

Audited by Gen Agent Trust Hub on Jun 13, 2026

Risk Level: SAFE
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill recommends installing official Google client libraries (google-genai and @google/genai) for interacting with the Gemini API. These are established, trusted packages from a well-known service provider.\n- [COMMAND_EXECUTION]: Provides instructions for using gcloud and curl to authenticate and interact with Google Cloud Platform services. These operations target official API endpoints (aiplatform.googleapis.com) and follow standard cloud resource management practices.\n- [PROMPT_INJECTION]: Contains directives for the agent (labeled as Mandatory Instructions) to enforce the use of specific model versions and SDKs in its responses. These serve as technical guardrails to ensure correct API usage. Additionally, the skill provides a bridge to an external API, creating a surface for indirect prompt injection. Ingestion points: User prompt strings interpolated into interaction requests. Boundary markers: None explicitly shown. Capability inventory: Network communication with the Gemini Interactions API. Sanitization: No explicit input filtering demonstrated in code snippets.\n- [DATA_EXFILTRATION]: Demonstrates how to handle authentication tokens using the gcloud auth print-access-token command. This approach follows security best practices by generating short-lived tokens at runtime rather than hardcoding sensitive credentials.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 13, 2026, 06:03 AM
Security Audit — agent-trust-hub — gemini-interactions-api